This paper presents a new Markov model to study travellers' stochastic behaviour in their day-to-day route choice adjustment process. The model is characterized by two components: how often a traveller reconsiders his/herroute choice (route switching rate), and what the probability is to take a certain route (route choice probability). By applying the evolutionary game theory, the conventional perfect information and complete rationality requirements in equilibrium analysis are relaxed. A deterministic mean (expected) route flow dynamic is derived which closely approximates the underlying route flow stochastic process in any finite time span as the travel demand grows large. The mean dynamic is general in that many existing deterministic processes can be considered as its special cases, and more importantly, their meaningful individual behaviour explanations are unveiled. It can be shown that with certain reasonable assumptions of behavioural rules of route-switching rate and route choice probability, the Wardrop userequilibrium can be approached by travellers' day-to-day behaviour adjustment process, even if an individual traveller only has access to incompleteinformation and exhibits limited rationality. In addition, the day-to-daymean route flow dynamic may evolve to user equilibrium, the stochastic user equilibrium, system optimal and other disequilibrium states depending on different behavioural rules of route-switching rate and route choice probability, network supply and congestion toll pricing schemes. This model is particularly useful to study the resulting day-to-day disequilibrium traffic evolving pattern when a portion of a network structure is to undergo a scheduled upgrade or when a capacity reduction takes place due to external interventions. This was found to be the case with three testing scenarios. For the covering abstract see ITRD E144727. Reprinted with permissionof Elsevier.
Abstract